{"id":"https://openalex.org/W7137936492","doi":"https://doi.org/10.48550/arxiv.2603.15600","title":"From Passive Observer to Active Critic: Reinforcement Learning Elicits Process Reasoning for Robotic Manipulation","display_name":"From Passive Observer to Active Critic: Reinforcement Learning Elicits Process Reasoning for Robotic Manipulation","publication_year":2026,"publication_date":"2026-03-16","ids":{"openalex":"https://openalex.org/W7137936492","doi":"https://doi.org/10.48550/arxiv.2603.15600"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2603.15600","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.15600","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"article"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2603.15600","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5100767941","display_name":"Yibin Liu","orcid":"https://orcid.org/0000-0001-5021-9412"},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Liu, Yibin","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5006651724","display_name":"Yaxing Lyu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lyu, Yaxing","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129687709","display_name":"Daqi Gao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Gao, Daqi","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129735125","display_name":"Zhixuan Liang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liang, Zhixuan","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129688023","display_name":"Weiliang Tang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tang, Weiliang","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5053302545","display_name":"Shilong Mu","orcid":"https://orcid.org/0009-0004-3638-6539"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Mu, Shilong","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129671707","display_name":"Xiaokang Yang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yang, Xiaokang","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5129677490","display_name":"Yao Mu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Mu, Yao","raw_affiliation_strings":[],"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":8,"corresponding_author_ids":["https://openalex.org/A5100767941"],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10812","display_name":"Human Pose and Action Recognition","score":0.3391000032424927,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T10812","display_name":"Human Pose and Action Recognition","score":0.3391000032424927,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10462","display_name":"Reinforcement Learning in Robotics","score":0.10620000213384628,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10653","display_name":"Robot Manipulation and Learning","score":0.08269999921321869,"subfield":{"id":"https://openalex.org/subfields/2207","display_name":"Control and Systems Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/reinforcement-learning","display_name":"Reinforcement learning","score":0.7049000263214111},{"id":"https://openalex.org/keywords/leverage","display_name":"Leverage (statistics)","score":0.6432999968528748},{"id":"https://openalex.org/keywords/bottleneck","display_name":"Bottleneck","score":0.5130000114440918},{"id":"https://openalex.org/keywords/benchmark","display_name":"Benchmark (surveying)","score":0.4959999918937683},{"id":"https://openalex.org/keywords/process","display_name":"Process (computing)","score":0.4749000072479248},{"id":"https://openalex.org/keywords/generalization","display_name":"Generalization","score":0.459199994802475},{"id":"https://openalex.org/keywords/observer","display_name":"Observer (physics)","score":0.45829999446868896},{"id":"https://openalex.org/keywords/task","display_name":"Task (project management)","score":0.4514000117778778}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7373999953269958},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.7049000263214111},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.6689000129699707},{"id":"https://openalex.org/C153083717","wikidata":"https://www.wikidata.org/wiki/Q6535263","display_name":"Leverage (statistics)","level":2,"score":0.6432999968528748},{"id":"https://openalex.org/C2780513914","wikidata":"https://www.wikidata.org/wiki/Q18210350","display_name":"Bottleneck","level":2,"score":0.5130000114440918},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.4982999861240387},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.4959999918937683},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.4749000072479248},{"id":"https://openalex.org/C177148314","wikidata":"https://www.wikidata.org/wiki/Q170084","display_name":"Generalization","level":2,"score":0.459199994802475},{"id":"https://openalex.org/C2780704645","wikidata":"https://www.wikidata.org/wiki/Q9251458","display_name":"Observer (physics)","level":2,"score":0.45829999446868896},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.4514000117778778},{"id":"https://openalex.org/C48103436","wikidata":"https://www.wikidata.org/wiki/Q599031","display_name":"State (computer science)","level":2,"score":0.4088999927043915},{"id":"https://openalex.org/C60692881","wikidata":"https://www.wikidata.org/wiki/Q584529","display_name":"Humanoid robot","level":3,"score":0.39579999446868896},{"id":"https://openalex.org/C34413123","wikidata":"https://www.wikidata.org/wiki/Q170978","display_name":"Robotics","level":3,"score":0.32249999046325684},{"id":"https://openalex.org/C150415221","wikidata":"https://www.wikidata.org/wiki/Q40687","display_name":"Robotic arm","level":2,"score":0.3140000104904175},{"id":"https://openalex.org/C14036430","wikidata":"https://www.wikidata.org/wiki/Q3736076","display_name":"Function (biology)","level":2,"score":0.3057999908924103},{"id":"https://openalex.org/C196340769","wikidata":"https://www.wikidata.org/wiki/Q7698910","display_name":"Temporal difference learning","level":3,"score":0.30059999227523804},{"id":"https://openalex.org/C175154964","wikidata":"https://www.wikidata.org/wiki/Q380077","display_name":"Task analysis","level":3,"score":0.3005000054836273},{"id":"https://openalex.org/C90509273","wikidata":"https://www.wikidata.org/wiki/Q11012","display_name":"Robot","level":2,"score":0.27639999985694885},{"id":"https://openalex.org/C136389625","wikidata":"https://www.wikidata.org/wiki/Q334384","display_name":"Supervised learning","level":3,"score":0.27480000257492065},{"id":"https://openalex.org/C2775924081","wikidata":"https://www.wikidata.org/wiki/Q55608371","display_name":"Control (management)","level":2,"score":0.2727000117301941},{"id":"https://openalex.org/C2779903281","wikidata":"https://www.wikidata.org/wiki/Q6888026","display_name":"Modalities","level":2,"score":0.2678000032901764},{"id":"https://openalex.org/C2984634286","wikidata":"https://www.wikidata.org/wiki/Q1331926","display_name":"Decision process","level":2,"score":0.2515000104904175}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2603.15600","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.15600","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"article"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2603.15600","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.15600","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"article"},"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/16","score":0.5218331217765808,"display_name":"Peace, Justice and strong institutions"}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Accurate":[0],"process":[1],"supervision":[2],"remains":[3],"a":[4,22,58,85,132],"critical":[5],"challenge":[6],"for":[7,78],"long-horizon":[8],"robotic":[9],"manipulation.":[10],"A":[11],"primary":[12],"bottleneck":[13],"is":[14],"that":[15,31,61,121],"current":[16,39,98],"video":[17,63,93],"MLLMs,":[18],"trained":[19],"primarily":[20],"under":[21],"Supervised":[23],"Fine-Tuning":[24],"(SFT)":[25],"paradigm,":[26],"function":[27],"as":[28],"passive":[29],"\"Observers\"":[30],"recognize":[32],"ongoing":[33],"events":[34],"rather":[35],"than":[36],"evaluating":[37],"the":[38,43,92,103,136],"state":[40,99],"relative":[41,146],"to":[42,73],"final":[44],"task":[45],"goal.":[46],"In":[47],"this":[48],"paper,":[49],"we":[50],"introduce":[51],"PRIMO":[52,105,122,154],"R1":[53,123,155],"(Process":[54],"Reasoning":[55],"Induced":[56],"Monitoring),":[57],"7B":[59,129],"framework":[60],"transforms":[62],"MLLMs":[64],"into":[65],"active":[66],"\"Critics\".":[67],"We":[68,165],"leverage":[69],"outcome-based":[70],"Reinforcement":[71],"Learning":[72],"incentivize":[74],"explicit":[75],"Chain-of-Thought":[76],"generation":[77],"progress":[79],"estimation.":[80],"Furthermore,":[81,153],"our":[82,128],"architecture":[83],"constructs":[84],"structured":[86],"temporal":[87],"input":[88],"by":[89,102,181],"explicitly":[90],"anchoring":[91],"sequence":[94],"between":[95],"initial":[96],"and":[97,107,115],"images.":[100],"Supported":[101],"proposed":[104],"Dataset":[106],"Benchmark,":[108],"extensive":[109],"experiments":[110],"across":[111],"diverse":[112],"in-domain":[113],"environments":[114],"out-of-domain":[116],"real-world":[117],"humanoid":[118],"scenarios":[119],"demonstrate":[120],"achieves":[124,131],"state-of-the-art":[125,167],"performance.":[126],"Quantitatively,":[127],"model":[130],"50%":[133],"reduction":[134],"in":[135],"mean":[137],"absolute":[138],"error":[139],"of":[140],"specialized":[141],"reasoning":[142],"baselines,":[143],"demonstrating":[144],"significant":[145],"accuracy":[147],"improvements":[148],"over":[149],"72B-scale":[150],"general":[151],"MLLMs.":[152],"exhibits":[156],"strong":[157],"zero-shot":[158],"generalization":[159],"on":[160,169],"difficult":[161],"failure":[162],"detection":[163],"tasks.":[164],"establish":[166],"performance":[168],"RoboFail":[170],"benchmark":[171],"with":[172],"67.0%":[173],"accuracy,":[174],"surpassing":[175],"closed-source":[176],"models":[177],"like":[178],"OpenAI":[179],"o1":[180],"6.0%.":[182]},"counts_by_year":[],"updated_date":"2026-03-18T06:31:55.123368","created_date":"2026-03-18T00:00:00"}